Measurement reports on individual wildfires: what our satellites saw, when, and what they missed. Every figure traces back to single detections with timestamp, source and measured value.
On 13 and 14 August 2026, two wildfires broke out within a single day on the Belgian-German border, just over thirty kilometres apart. Both became a record for their country. We hold 4,127 satellite detections of them. This is what that series says about speed, progression, and the limits of what satellites can see.
The fire in the Hautes Fagnes started on the afternoon of Friday 14 August at Fagne des Deux-Séries in the municipality of Baelen, reportedly at around a quarter past one. From one hectare it grew in just over a day to roughly 2,700 hectares and then to about 3,000, surpassing the record set by the 2011 peat fire in the same area.
The province of Liège declared a disaster phase. In Sourbrodt emergency services went door to door in some fifteen streets; in nearby Bütgenbach residents of several streets had to leave, about 600 people in total. Tourists were advised to move to safer ground or go home. The smoke was visible as far as the southern Netherlands, and police in Trier warned of smoke plumes over the Eifel and the Moselle.
Geostationary satellites watch the same part of Europe continuously and deliver an image every ten minutes. That is why the first bar here is also the earliest one that could have existed. Height and colour follow the radiated power. Note the first bar: 13:00 is the nominal start of the scan slot rather than an exact moment, and the fire was reported at around a quarter past one. Detection and report therefore fall within the same quarter of an hour.
Eight satellite sources saw this fire. What matters is not when a satellite passed over, but when its image had been processed and was usable. Measured from 13:00, the moment of the first detection.
| Source | Acquired | Available | After first detection |
|---|---|---|---|
| MTG FRP-PIXELgeostationary | 13:00 | 13:34 | 35 min |
| MSG FRP-PIXELgeostationary | 14:45 | 15:22 | 142 min |
| MODISpolar | 15:29 | 16:33 | 214 min |
| VIIRS NOAA-20polar | 14:21 | 16:55 | 235 min |
| VIIRS Suomi-NPPpolar | 14:02 | 17:16 | 257 min |
| VIIRS NOAA-21polar | 15:06 | 17:38 | 279 min |
| Sentinel-3Apolar | 22:28 | 01:05 | +726 min |
| Sentinel-3Bpolar | 23:31 | 02:10 | +790 min |
Sentinel-3 is included for completeness. That source is valuable for mapping a burnt area, not for the first warning.
The difference is not a matter of better software. Geostationary satellites hover over Europe; polar satellites pass over a few times a day and have to play their data back to a ground station first. Anyone relying on polar sources alone is structurally three to five hours late for an afternoon fire.
Detections per hour, local time. The dotted line marks the highest power measured across the whole fire: 1,355 MW on the morning of 15 August. A peat fire staying lit through the night is typical; ordinary forest fires fall back much further between afternoons.
For a fire that grows fast across open heath and peat, geostationary is decisive: MTG saw it two to three hours before any polar satellite. Compare this with our report on the Hürtgenwald a day earlier, where it went exactly the other way.
From 17 August 01:40 onwards we recorded no further detections for this fire, even though it was not yet under control at that point.
That is not a malfunction but the physics of the measurement. Satellites measure heat radiated at the surface. Burning peat smoulders on below ground, sometimes for weeks, without giving off enough heat to register as an anomaly. That is precisely why such a fire can resurface days later somewhere else.
Satellite detection is therefore strong at the earliest moment and on flaring fire, and weak on a smouldering subsurface. Presenting it the other way round means selling something that does not exist.